First experimental implementation of the Clear-Site-Data header

content/browser/frame_host/navigation_handle_impl.cc

Index: content/browser/frame_host/navigation_handle_impl.cc
diff --git a/content/browser/frame_host/navigation_handle_impl.cc b/content/browser/frame_host/navigation_handle_impl.cc
index da20a1ee15fb52f76dc45c63fd26058e4f232110..520f0a4102e944ed2fd7fa34068e4ac48294615e 100644
--- a/content/browser/frame_host/navigation_handle_impl.cc
+++ b/content/browser/frame_host/navigation_handle_impl.cc
@@ -6,7 +6,9 @@
 
 #include <utility>
 
+#include "base/command_line.h"
 #include "base/logging.h"
+#include "content/browser/browsing_data/clear_site_data_throttle.h"
 #include "content/browser/devtools/render_frame_devtools_agent_host.h"
 #include "content/browser/frame_host/frame_tree_node.h"
 #include "content/browser/frame_host/navigator.h"
@@ -17,6 +19,7 @@
 #include "content/public/browser/content_browser_client.h"
 #include "content/public/common/browser_side_navigation_policy.h"
 #include "content/public/common/content_client.h"
+#include "content/public/common/content_switches.h"
 #include "net/url_request/redirect_info.h"
 #include "url/gurl.h"
 #include "url/url_constants.h"
@@ -31,6 +34,11 @@ void UpdateThrottleCheckResult(
   *to_update = result;
 }
 
+bool AreExperimentalFeaturesEnabled() {
+  return base::CommandLine::ForCurrentProcess()->HasSwitch(
+      switches::kEnableExperimentalWebPlatformFeatures);
+}
+
 }  // namespace
 
 // static
@@ -525,7 +533,14 @@ void NavigationHandleImpl::RunCompleteCallback(
 }
 
 void NavigationHandleImpl::RegisterNavigationThrottles() {
-  // Register the navigation throttles. The ScopedVector returned by
+  // Register the platform's navigation throttles.
+  if (AreExperimentalFeaturesEnabled()) {
+    std::unique_ptr<content::NavigationThrottle> clear_site_data_throttle =
+        content::ClearSiteDataThrottle::CreateThrottleFor(this);
+    throttles_.push_back(std::move(clear_site_data_throttle));

[clamy, 2016/07/18 14:47:49]

If the throttle is put there, it means it will execute before other throttles
that can cancel the navigation (potentially for security reasons). I'm not sure
this is the behavior we want.

Also considering that the throttle doesn't actually pause or cancel the
navigation, can we add a TODO to move this to a WebContentsObserver once we call
WCO::ReadyToCommitNavigation in the current architecture (see crbug.com/621856).
I don't see a good reason to have it as a throttle once we get the WCO call
after the response.

[msramek, 2016/07/18 17:03:20]

Done.

Good point - moved it to the end.

We haven't yet decided whether the data should be cleared synchronously or not.
The former might be a horrible user experience, the latter can lead to wrong
assumptions or unexpected behavior on the website if it starts writing data
before clearing was finished. We started with the asynchronous version, as it's
simpler and we want to see the how it works in the wild.

It is possible that will yet decide that we want synchronous clearing, in which
case the throttle should delay navigation. I mentioned that in the TODO.

+  }
+
+  // Register the embedder's navigation throttles. The ScopedVector returned by
   // GetNavigationThrottles is not assigned to throttles_ directly because it
   // would overwrite any throttle previously added with
   // RegisterThrottleForTesting.